Load Libraries & Set Themes
library(devtools) # Reproducibility (see end of file)
library(phyloseq) # Easier data manipulation
library(tidyverse) # Pretty plotting and data manipulation
library(forcats) # Recoding factors
library(cowplot) # Multiple plotting
library(d3heatmap) # For nice heatmaps
library(phytools) # Working with phylogenetic trees
library(ggtree) # Pretty ggplot-esque phylogenetic trees
library(ape) # Working with phylogenetic trees
library(DT) # Pretty Tables
library(gplots) # Heatmap.2 function
library(stringr) # To replace zeros in OTU names
source("code/set_colors.R") # Set Colors for plotting
# Set some global ggplot parameters
mytheme <- theme(legend.text = element_text(size = 8), legend.title = element_text(size = 8, face = "bold"),
plot.title = element_text(size = 10), legend.position = c(0.01, 0.9),
axis.title = element_text(size = 10, face = "bold"), axis.text = element_text(size = 8),
legend.key.width=unit(0.1,"line"), legend.key.height=unit(0.1,"line"))
Load FCM Data
# Read in the data
raw_data <- read.table(file="data/Chloroplasts_removed/old/nochloro_HNA_LNA.tsv", header = TRUE) %>%
mutate(norep_filter_name = paste(substr(Sample_16S,1,4), substr(Sample_16S,6,9), sep = "")) %>%
arrange(norep_filter_name)
# Subset out only the Muskegon and Surface samples
muskegon_data <- raw_data %>%
dplyr::filter(Lake == "Muskegon" & Depth == "Surface") %>%
dplyr::select(norep_filter_name)
# Load in the productivity data
production <- read.csv(file="data/production_data.csv", header = TRUE) %>%
dplyr::filter(fraction == "Free" & limnion == "Surface") %>% # Select only rows that are free-living
dplyr::select(names, tot_bacprod, SD_tot_bacprod) %>% # Select relevant columns for total productivity
mutate(tot_bacprod = round(tot_bacprod, digits = 2), # Round to 2 decimals
SD_tot_bacprod = round(SD_tot_bacprod, digits = 2) ) %>%
dplyr::rename(norep_filter_name = names) %>% # Rename to match other data frame
arrange(norep_filter_name)
# Stop if the names do not match
stopifnot(muskegon_data$norep_filter_name == production$norep_filter_name)
# Combine the two muskegon data frames into one
combined_data <- left_join(muskegon_data, production, by = "norep_filter_name")
# Merge the combined data back into the original data frame (data)
data <- left_join(raw_data, combined_data, by = "norep_filter_name") %>%
dplyr::select(-c(Platform, Fraction)) %>%
mutate(Lake = factor(Lake, levels = c("Michigan","Inland", "Muskegon")))
head(data)
## samples Total.cells HNA.cells LNA.cells Total.count.sd HNA.sd LNA.sd Lake Sample_16S Season Month Year Site Depth Total_Sequences norep_filter_name tot_bacprod SD_tot_bacprod
## 1 110D2-115-2 620420.9 139880.7 480540.2 1640.275 4639.952 5795.307 Michigan 110D2F115 Winter Januari 2015 MM110 Deep 19654 110DF115 NA NA
## 2 110D2-415-2 799314.0 155198.1 644115.9 9388.845 4198.741 5493.442 Michigan 110D2F415 Spring April 2015 MM110 Deep 7951 110DF415 NA NA
## 3 110D2-515 1261369.4 362443.6 898925.9 15341.779 10504.696 6299.495 Michigan 110D2F515 Spring May 2015 MM110 Deep 16293 110DF515 NA NA
## 4 110D2-715-2 542723.6 131132.1 411591.5 6139.696 2196.008 3967.958 Michigan 110D2F715 Summer July 2015 MM110 Deep 16882 110DF715 NA NA
## 5 110D2-815 548027.9 131820.5 416207.4 3552.010 3805.294 2441.935 Michigan 110D2F815 Summer August 2015 MM110 Deep 22157 110DF815 NA NA
## 6 110D2-915-2 731931.0 189746.2 542184.8 36726.396 6634.230 30473.925 Michigan 110D2F915 Fall September 2015 MM110 Deep 16500 110DF915 NA NA
# Fix the metadata
data$Month[data$Month == "Januari"] <- "June"
data$Season[data$Season == "Winter"] <- "Summer"
# Write out the file
#write.table(data, file="data/Chloroplasts_removed/productivity_data.tsv", row.names=TRUE)
####### HELPFUL TRANSFORMED AND SUBSETTED DATA
##### Subset Muskegon Lake Data Only
muskegon <- dplyr::filter(data, Lake == "Muskegon" & Depth == "Surface") %>%
mutate(Site = factor(Site, levels = c("MOT", "MDP", "MBR", "MIN"))) %>%
filter(tot_bacprod < 90)
# Gather the data for summary statistics
df_cells <- data %>%
dplyr::select(1:4, Lake, Season:Depth) %>%
rename(Total = Total.cells, HNA = HNA.cells, LNA = LNA.cells) %>%
gather(key = FCM_type, value = num_cells, Total:LNA)
Summary & Descriptive Statistics
# Number of cells in HNA and LNA across all samples
# For a plot of these values, see figure 1 below
df_cells %>%
group_by(FCM_type) %>%
summarise(num_samples = n(),
mean_cells = round(mean(num_cells), digits = 2),
sd_mean_cells = round(sd(num_cells), digits = 2),
median_cells = round(median(num_cells), digits = 2)) %>%
datatable(caption = "Mean and median number of cells per ecosystem", rownames = FALSE)
# Are there more total cells in one lake over the other?
totcells_df <- df_cells %>%
filter(FCM_type == "Total")
# Compute the analysis of variance
totcells_aov <- aov(num_cells ~ Lake, data = totcells_df)
summary(totcells_aov) # there is a difference, but which lake?
## Df Sum Sq Mean Sq F value Pr(>F)
## Lake 2 4.332e+14 2.166e+14 37.75 2.72e-14 ***
## Residuals 170 9.755e+14 5.738e+12
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Which samples are significant from each other?
TukeyHSD(totcells_aov) # Michigan is significantly different form Muskegon and Inland
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = num_cells ~ Lake, data = totcells_df)
##
## $Lake
## diff lwr upr p adj
## Inland-Michigan 3417194.2 2411737 4422651.4 0.0000000
## Muskegon-Michigan 3030294.1 1939004 4121584.0 0.0000000
## Muskegon-Inland -386900.1 -1489077 715276.9 0.6849527
# Number of cells in HNA and LNA per ecosystem
df_cells %>%
group_by(Lake, FCM_type) %>%
summarise(num_samples = n(),
mean_cells = round(mean(num_cells), digits = 2),
median_cells = round(median(num_cells), digits = 2)) %>%
datatable(caption = "Mean and median number of cells per ecosystem", rownames = FALSE)
# Proportion of HNA and LNA across all samples
df_cells %>%
dplyr::select(samples, Lake, FCM_type, num_cells) %>%
spread(FCM_type, num_cells) %>%
mutate(prop_HNA = HNA/Total * 100,
prop_LNA = LNA/Total * 100) %>%
dplyr::select(Lake, prop_HNA, prop_LNA) %>%
summarize(mean_HNA = round(mean(prop_HNA), digits = 2),
sd_HNA = round(sd(prop_HNA), digits = 2),
mean_LNA = round(mean(prop_LNA), digits = 2),
sd_LNA = round(sd(prop_LNA), digits = 2))
## mean_HNA sd_HNA mean_LNA sd_LNA
## 1 30.41 9.06 69.59 9.06
# Proportion of HNA and LNAper ecosystem
prop_stats <- df_cells %>%
dplyr::select(samples, Lake, FCM_type, num_cells) %>%
spread(FCM_type, num_cells) %>%
mutate(prop_HNA = HNA/Total * 100,
prop_LNA = LNA/Total * 100) %>%
dplyr::select(Lake, prop_HNA, prop_LNA) %>%
rename(HNA = prop_HNA, LNA = prop_LNA) %>%
group_by(Lake) %>%
summarize(mean_HNA = round(mean(HNA), digits = 2),
mean_LNA = round(mean(LNA), digits = 2),
min_HNA = round(min(HNA), digits = 2),
max_HNA = round(max(HNA), digits = 2),
min_LNA = round(min(LNA), digits = 2),
max_LNA = round(max(LNA), digits = 2),
num_samples = n())
datatable(prop_stats, caption = "Statistics of the percentage of each flow cytometry group across the systems", rownames = FALSE)
Figure 1
# Load in the data from each lake
musk_all_df <- read.table(file="data/Chloroplasts_removed/ByLake_Filtering/5in10/muskegon/muskegon_sampledata_5in10.tsv", header = TRUE) %>%
mutate(norep_filter_name = paste(substr(Sample_16S,1,4), substr(Sample_16S,6,9), sep = "")) %>%
arrange(norep_filter_name)
mich_all_df <- read.table(file="data/Chloroplasts_removed/ByLake_Filtering/5in10/michigan/michigan_sampledata_5in10.tsv", header = TRUE) %>%
mutate(norep_filter_name = paste(substr(Sample_16S,1,4), substr(Sample_16S,6,9), sep = "")) %>%
arrange(norep_filter_name)
inla_all_df <- read.table(file="data/Chloroplasts_removed/ByLake_Filtering/5in10/inland/inland_sampledata_5in10.tsv", header = TRUE) %>%
mutate(norep_filter_name = paste(substr(Sample_16S,1,4), substr(Sample_16S,6,9), sep = "")) %>%
arrange(norep_filter_name)
stopifnot(colnames(musk_all_df) == colnames(mich_all_df))
stopifnot(colnames(musk_all_df) == colnames(inla_all_df))
lakes <- bind_rows(musk_all_df, mich_all_df, inla_all_df)
p1 <- ggplot(df_cells, aes(x = FCM_type, y = num_cells, fill = Lake, color = Lake, shape = Lake)) +
geom_point(size = 1.5, position = position_jitterdodge(), color = "black") +
geom_boxplot(alpha = 0.7, outlier.shape = NA, show.legend = FALSE, color = "black") +
scale_color_manual(values = lake_colors, guide = "none") +
scale_fill_manual(values = lake_colors) +
scale_shape_manual(values = lake_shapes) +
labs(y = "Number of Cells (cells/mL)", x = "FCM Type") +
mytheme + theme(legend.title = element_blank(), legend.position = c(0.01, 0.95)) +
guides(colour = guide_legend(override.aes = list(size=2.5)),
shape = guide_legend(override.aes = list(size=2.5)),
fill = guide_legend(override.aes = list(size=2.5)))
## Is there a corrlation between HNA and LNA across ecosystems?
# 1. Run the linear model
lm_allNA_corr <- lm(LNA.cells ~ HNA.cells, data = lakes)
summary(lm(LNA.cells ~ HNA.cells * Lake, data = lakes))
##
## Call:
## lm(formula = LNA.cells ~ HNA.cells * Lake, data = lakes)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3651006 -558525 -113725 511977 3997810
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.699e+06 3.590e+05 10.304 < 2e-16 ***
## HNA.cells 3.827e-01 1.704e-01 2.246 0.026 *
## LakeMichigan -2.995e+06 4.523e+05 -6.622 4.63e-10 ***
## LakeMuskegon -2.363e+06 5.411e+05 -4.367 2.21e-05 ***
## HNA.cells:LakeMichigan 5.404e-01 4.257e-01 1.269 0.206
## HNA.cells:LakeMuskegon 1.027e+00 2.549e-01 4.029 8.50e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1301000 on 167 degrees of freedom
## Multiple R-squared: 0.6116, Adjusted R-squared: 0.6
## F-statistic: 52.6 on 5 and 167 DF, p-value: < 2.2e-16
## 2. Extract the R2 and p-value from the linear model:
lm_allNA_corr_stats <- paste("atop(R^2 ==", round(summary(lm_allNA_corr)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_allNA_corr)$coefficients[,4][2]), digits = 24), ")")
# 3. Plot it
p2 <- ggplot(lakes, aes(x = HNA.cells, y = LNA.cells)) +
geom_errorbarh(aes(xmin = HNA.cells - HNA.sd, xmax = HNA.cells + HNA.sd), color = "grey", alpha = 0.8) +
geom_errorbar(aes(ymin = LNA.cells - LNA.sd, max = LNA.cells + LNA.sd), color = "grey", alpha = 0.8) +
geom_point(size = 2.5, alpha = 0.9, aes(fill = Lake, shape = Lake)) +
scale_fill_manual(values = lake_colors) +
scale_shape_manual(values = lake_shapes) +
geom_smooth(method = "lm", color = "black") +
labs(y = "LNA Cell Density (cells/mL)", x = "HNA Cell Density (cells/mL)") +
scale_x_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 6.1e+06)) +
scale_y_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 1e+07)) +
annotate("text", x=5e+06, y=0.8e+06, label=lm_allNA_corr_stats, parse = TRUE, color = "black", size = 3) +
mytheme + theme(legend.position = "none") #theme(legend.title = element_blank(), legend.position = c(0.01, 0.95))
# Figure 1
plot_grid(p1, p2, ncol = 2, nrow =1, labels = c("A","B"), rel_widths = c(0.95, 1))
Figure 1B: Plotted By Lake System
### MUSKEGON ONLY ANALYSIS
# 1. Run the linear model
lm_muskNA_corr <- lm(LNA.cells ~ HNA.cells, data = musk_all_df)
## 2. Extract the R2 and p-value from the linear model:
lm_muskNA_corr_stats <- paste("atop(R^2 ==", round(summary(lm_muskNA_corr)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_muskNA_corr)$coefficients[,4][2]), digits = 9), ")")
musk_corr_plot <- ggplot(musk_all_df, aes(x = HNA.cells, y = LNA.cells)) +
geom_errorbarh(aes(xmin = HNA.cells - HNA.sd, xmax = HNA.cells + HNA.sd), color = "grey") +
geom_errorbar(aes(ymin = LNA.cells - LNA.sd, max = LNA.cells + LNA.sd), color = "grey") +
geom_point(size = 3, shape = 22, aes(fill = Lake)) +
geom_smooth(method = "lm", color = "black") +
ggtitle("Muskegon Lake") + scale_fill_manual(values = lake_colors) +
scale_x_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 6.1e+06)) +
scale_y_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 1e+07)) +
labs(y = "LNA Cell Density (cells/mL)", x = "HNA Cell Density (cells/mL)") +
annotate("text", x=5e+06, y=0.8e+06, label=lm_muskNA_corr_stats, parse = TRUE, color = "black", size = 3) +
mytheme
### INLAND ONLY ANALYSIS
# 1. Run the linear model
lm_inlaNA_corr <- lm(LNA.cells ~ HNA.cells, data = filter(inla_all_df, Sample_16S != "Z14003F"))
## 2. Extract the R2 and p-value from the linear model:
lm_inlaNA_corr_stats <- paste("atop(R^2 ==", round(summary(lm_inlaNA_corr)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_inlaNA_corr)$coefficients[,4][2]), digits = 2), ")")
inla_corr_plot <- ggplot(filter(inla_all_df, Sample_16S != "Z14003F"), aes(x = HNA.cells, y = LNA.cells)) +
geom_errorbarh(aes(xmin = HNA.cells - HNA.sd, xmax = HNA.cells + HNA.sd), color = "grey") +
geom_errorbar(aes(ymin = LNA.cells - LNA.sd, max = LNA.cells + LNA.sd), color = "grey") +
geom_point(size = 3, shape = 22, aes(fill = Lake)) +
geom_smooth(method = "lm", color = "black") +
scale_x_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 6.1e+06)) +
scale_y_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 1e+07)) +
ggtitle("Inland Lakes") + scale_fill_manual(values = lake_colors) +
labs(y = "LNA Cell Density (cells/mL)", x = "HNA Cell Density (cells/mL)") +
annotate("text", x=5e+06, y=0.8e+06, label=lm_inlaNA_corr_stats, parse = TRUE, color = "black", size = 3) +
mytheme
### MICHIGAN ONLY ANALYSIS
# 1. Run the linear model
lm_michNA_corr <- lm(LNA.cells ~ HNA.cells, data = mich_all_df)
# Without the Lake Michigan outlier!
summary(lm(LNA.cells ~ HNA.cells, data = filter(mich_all_df, Sample_16S != "M15S2F515")))
##
## Call:
## lm(formula = LNA.cells ~ HNA.cells, data = filter(mich_all_df,
## Sample_16S != "M15S2F515"))
##
## Residuals:
## Min 1Q Median 3Q Max
## -599787 -239620 -95094 241831 987737
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 5.832e+05 9.793e+04 5.956 3.37e-07 ***
## HNA.cells 1.200e+00 1.797e-01 6.678 2.78e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 354600 on 46 degrees of freedom
## Multiple R-squared: 0.4922, Adjusted R-squared: 0.4812
## F-statistic: 44.59 on 1 and 46 DF, p-value: 2.778e-08
## 2. Extract the R2 and p-value from the linear model:
lm_michNA_corr_stats <- paste("atop(R^2 ==", round(summary(lm_michNA_corr)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_michNA_corr)$coefficients[,4][2]), digits = 11), ")")
mich_corr_plot <- ggplot(mich_all_df, aes(x = HNA.cells, y = LNA.cells)) +
geom_errorbarh(aes(xmin = HNA.cells - HNA.sd, xmax = HNA.cells + HNA.sd), color = "grey") +
geom_errorbar(aes(ymin = LNA.cells - LNA.sd, max = LNA.cells + LNA.sd), color = "grey") +
geom_point(size = 3, shape = 22, aes(fill = Lake)) +
geom_smooth(method = "lm", color = "black") +
scale_x_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 6.1e+06)) +
scale_y_continuous(labels = function(x) format(x, scientific = TRUE), limits = c(0, 1e+07)) +
ggtitle("Lake Michigan") + scale_fill_manual(values = lake_colors) +
labs(y = "LNA Cell Density (cells/mL)", x = "HNA Cell Density (cells/mL)") +
annotate("text", x=5e+06, y=0.8e+06, label=lm_michNA_corr_stats, parse = TRUE, color = "black", size = 3) +
mytheme
plot_grid(mich_corr_plot, inla_corr_plot, musk_corr_plot,
labels = c("A", "B", "C"), nrow = 1, ncol = 3)
Figure 2: FCM vs Productivity
######################## Analysis of HNA/LNA/Total Cells vs Total Productivity
# 1. Run the linear model
lm_HNA <- lm(tot_bacprod ~ HNA.cells, data = muskegon)
## 2. Extract the R2 and p-value from the linear model:
lm_HNA_stats <- paste("atop(R^2 ==", round(summary(lm_HNA)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_HNA)$coefficients[,4][2]), digits = 5), ")")
# 3. Plot it
HNA_vs_prod <- ggplot(muskegon, aes(x = HNA.cells, y = tot_bacprod)) +
geom_errorbarh(aes(xmin = HNA.cells - HNA.sd, xmax = HNA.cells + HNA.sd), color = "grey", alpha = 0.8) +
geom_errorbar(aes(ymin = tot_bacprod - SD_tot_bacprod, max = tot_bacprod + SD_tot_bacprod), color = "grey", alpha = 0.8) +
geom_point(size = 2, shape = 22, fill = "deepskyblue4") +
geom_smooth(method = "lm", color = "deepskyblue4") +
labs(y = "Total Bacterial Production \n (μg C/L/day)", x = "HNA Cell Density \n(cells/mL)") +
scale_x_continuous(labels = function(x) format(x, scientific = TRUE),
breaks = c(2e+06, 3e+06)) +
annotate("text", x=1.65e+06, y=60, label=lm_HNA_stats, parse = TRUE, color = "black", size = 3) +
mytheme
# 1. Run the linear model
lm_LNA <- lm(tot_bacprod ~ LNA.cells, data = muskegon)
## 2. Extract the R2 and p-value from the linear model:
lm_LNA_stats <- paste("atop(R^2 ==", round(summary(lm_LNA)$adj.r.squared, digits = 3), ",",
"p ==", round(unname(summary(lm_LNA)$coefficients[,4][2]), digits = 2), ")")
# 3. Plot it
LNA_vs_prod <- ggplot(muskegon, aes(x = LNA.cells, y = tot_bacprod)) +
geom_errorbarh(aes(xmin = LNA.cells - LNA.sd, xmax = LNA.cells + LNA.sd), color = "grey", alpha = 0.8) +
geom_errorbar(aes(ymin = tot_bacprod - SD_tot_bacprod, max = tot_bacprod + SD_tot_bacprod), color = "grey", alpha = 0.8) +
geom_point(size = 2.5, shape = 22, fill = "darkgoldenrod1") +
labs(y = "Total Bacterial Production \n (μg C/L/day)", x = "LNA Cell Density \n(cells/mL)") +
geom_smooth(method = "lm", se = FALSE, linetype = "longdash", color = "darkgoldenrod1") +
annotate("text", x=2.75e+06, y=60, label=lm_LNA_stats, parse = TRUE, color = "red", size = 3) +
mytheme
# 1. Run the linear model
lm_total <- lm(tot_bacprod ~ Total.cells, data = muskegon)
## 2. Extract the R2 and p-value from the linear model:
lm_total_stats <- paste("atop(R^2 ==", round(summary(lm_total)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_total)$coefficients[,4][2]), digits = 2), ")")
# 3. Plot it
Total_vs_prod <- ggplot(muskegon, aes(x = Total.cells, y = tot_bacprod)) +
geom_errorbarh(aes(xmin = Total.cells - Total.count.sd, xmax = Total.cells + Total.count.sd), color = "grey", alpha = 0.8) +
geom_errorbar(aes(ymin = tot_bacprod - SD_tot_bacprod, max = tot_bacprod + SD_tot_bacprod), color = "grey", alpha = 0.8) +
scale_shape_manual(values = lake_shapes) +
geom_point(size = 2.5, shape = 22, fill = "black") +
labs(y = "Total Bacterial Production \n (μg C/L/day)", x = "Total Cell Density \n(cells/mL)") +
geom_smooth(method = "lm", color = "black") +
#geom_smooth(method = "lm", se = FALSE, linetype = "longdash", color = "red") +
annotate("text", x=5.25e+06, y=60, label=lm_total_stats, parse = TRUE, color = "black", size = 3) +
mytheme
#### ONLY THE THREE PLOTS
# Put all three plots together into one
plot_grid(HNA_vs_prod + theme(legend.position = "none"),
LNA_vs_prod + theme(axis.title.y = element_blank(), legend.position = "none"),
Total_vs_prod + theme(axis.title.y = element_blank(), legend.position = "none"),
labels = c("A", "B", "C"),
ncol = 3, rel_widths = c(0.95, 0.8, 0.8))
Fraction HNA vs Productivity
## Plot the fraction of HNA
fmusk <- muskegon %>%
mutate(fHNA = HNA.cells/Total.cells,
fLNA = LNA.cells/Total.cells)
lm_fHNA <- lm(tot_bacprod ~ fHNA, data = fmusk)
## 2. Extract the R2 and p-value from the linear model:
lm_fHNA_stats <- paste("atop(R^2 ==", round(summary(lm_fHNA)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_fHNA)$coefficients[,4][2]), digits = 3), ")")
fHNA_vs_prod <- ggplot(fmusk, aes(x = fHNA, y = tot_bacprod)) +
geom_errorbar(aes(ymin = tot_bacprod - SD_tot_bacprod, max = tot_bacprod + SD_tot_bacprod), color = "grey") +
geom_point(size = 3, aes(shape = Lake), fill = "deepskyblue4") +
geom_smooth(method = "lm", linetype = "longdash", color = "deepskyblue4") +
scale_x_continuous(limits = c(0.15, 0.9), breaks = seq(0.2, 0.9, by = 0.2)) +
ylab("Bacterial Production") + xlab("Fraction HNA") +
scale_shape_manual(values = lake_shapes) +
annotate("text", x= 0.22, y=60, label=lm_fHNA_stats, parse = TRUE, color = "black", size = 3) +
mytheme + theme(legend.position = "none")
### LNA Fraction
## Plot the fraction of HNA
lm_fLNA <- lm(tot_bacprod ~ fLNA, data = fmusk)
## 2. Extract the R2 and p-value from the linear model:
lm_fLNA_stats <- paste("atop(R^2 ==", round(summary(lm_fLNA)$adj.r.squared, digits = 2), ",",
"p ==", round(unname(summary(lm_fLNA)$coefficients[,4][2]), digits = 3), ")")
fLNA_vs_prod <- ggplot(fmusk, aes(x = fLNA, y = tot_bacprod)) +
geom_errorbar(aes(ymin = tot_bacprod - SD_tot_bacprod, max = tot_bacprod + SD_tot_bacprod), color = "grey") +
geom_point(size = 3, aes(shape = Lake), fill = "darkgoldenrod1") +
geom_smooth(method = "lm", color = "darkgoldenrod1", fill = "darkgoldenrod1", linetype = "longdash") +
scale_x_continuous(limits = c(0.15, 0.9), breaks = seq(0.2, 0.9, by = 0.2)) +
ylab("Bacterial Production") + xlab("Fraction LNA") +
scale_shape_manual(values = lake_shapes) +
annotate("text", x= 0.22, y=60, label=lm_fLNA_stats, parse = TRUE, color = "black", size = 3) +
mytheme + theme(legend.position = "none")
plot_grid(HNA_vs_prod + theme(legend.position = "none"),
fHNA_vs_prod,
LNA_vs_prod, fLNA_vs_prod,
labels = c("A", "B", "C", "D"), ncol = 2, nrow = 2,
align = "h")
Alternative Figure 1:
first_row <- plot_grid(HNA_vs_prod + theme(legend.position = "none"),
LNA_vs_prod + theme(axis.title.y = element_blank(), legend.position = "none"),
Total_vs_prod + theme(axis.title.y = element_blank(), legend.position = "none"),
labels = c("A", "B", "C"),
ncol = 3, nrow = 1,
rel_widths = c(0.95, 0.8, 0.8))
plot_for_legend <-
ggplot(df_cells, aes(x = Lake, y = num_cells, fill = FCM_type, color = FCM_type)) +
geom_point(size = 1, shape = 22, position = position_jitterdodge()) +
scale_fill_manual(values = fcm_colors) +
scale_color_manual(values = fcm_colors, guide = "none") +
scale_shape_manual(values = 22) +
scale_y_continuous(labels = function(x) format(x, scientific = TRUE)) +
labs(y = "Number of Cells \n (cells/mL)", x = "Lake") +
theme(legend.position = "right",
legend.title = element_text(size = 12, face = "bold"),
legend.text = element_text(size = 12),
legend.key.width=unit(1,"line"),
legend.key.height=unit(1,"line")) +
guides(fill = guide_legend(override.aes = list(size=3.5)))
legend1 <- get_legend(plot_for_legend)
legend2 <- get_legend(p2 + theme(legend.position = "right",
legend.title = element_text(size = 12, face = "bold"),
legend.text = element_text(size = 12),
legend.key.width=unit(1,"line"),
legend.key.height=unit(1,"line")) +
guides(fill = guide_legend(override.aes = list(size=3.5))))
leg_positions <- plot_grid(legend1, legend2, nrow = 2, ncol = 1)
second_row <- plot_grid(NULL, p2, leg_positions,
labels = c("", "D", ""),
ncol = 3, nrow = 1,
rel_widths = c(0.5, 1, 0.5))
## PLOT THE FIGURE
plot_grid(first_row, second_row,
nrow = 2, ncol = 1)
# REPOSITION PLOTS FOR NEW ORDER OF TEXT
new_leg_positions <- plot_grid(legend2, legend1, nrow = 2, ncol = 1)
swap_row1 <- plot_grid(NULL, p2, new_leg_positions,
labels = c("", "A", ""),
ncol = 3, nrow = 1,
rel_widths = c(0.5, 1, 0.5))
swap_row2 <- plot_grid(HNA_vs_prod + theme(legend.position = "none"),
LNA_vs_prod + theme(axis.title.y = element_blank(), legend.position = "none"),
Total_vs_prod + theme(axis.title.y = element_blank(), legend.position = "none"),
labels = c("B", "C", "D"),
ncol = 3, nrow = 1,
rel_widths = c(0.95, 0.8, 0.8))
plot_grid(swap_row1, swap_row2,
nrow = 2, ncol = 1)
Figure 4: Heatmap
Load RL Ranking data
# Read in Data
dfHNA_musk <- read.csv("Final/FS_Scores/Muskegon_fs_scores_HNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.09) %>%
mutate(RL.ranking = 1/RL.ranking, Lake = "Muskegon")%>%
rename(OTU = X, RL.ranking.HNA = RL.ranking) %>%
dplyr::select(OTU, RL.ranking.HNA)
dfLNA_musk <- read.csv("Final/FS_Scores/Muskegon_fs_scores_LNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.09) %>%
mutate(RL.ranking = -1/RL.ranking, Lake = "Muskegon") %>%
rename(OTU = X, RL.ranking.LNA = RL.ranking) %>%
dplyr::select(OTU, RL.ranking.LNA)
dfHNA_inland <- read.csv("Final/FS_Scores/Inland_fs_scores_HNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.13) %>%
mutate(RL.ranking = 1/RL.ranking, Lake = "Inland")%>%
rename(OTU = X, RL.ranking.HNA = RL.ranking) %>%
dplyr::select(OTU, RL.ranking.HNA)
dfLNA_inland <- read.csv("Final/FS_Scores/Inland_fs_scores_LNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.108) %>%
mutate(RL.ranking = -1/RL.ranking, Lake = "Inland") %>%
rename(OTU = X, RL.ranking.LNA = RL.ranking) %>%
dplyr::select(OTU, RL.ranking.LNA)
dfHNA_michigan <- read.csv("Final/FS_Scores/Michigan_fs_scores_HNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.248) %>%
mutate(RL.ranking = 1/RL.ranking, Lake = "Michigan")%>%
rename(OTU = X, RL.ranking.HNA = RL.ranking) %>%
dplyr::select(OTU, RL.ranking.HNA)
dfLNA_michigan <- read.csv("Final/FS_Scores/Michigan_fs_scores_LNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.286) %>%
mutate(RL.ranking = -1/RL.ranking, Lake = "Michigan") %>%
rename(OTU = X, RL.ranking.LNA = RL.ranking) %>%
dplyr::select(OTU, RL.ranking.LNA)
# Combine HNA and LNA datasets together
muskegon_df <- full_join(dfHNA_musk, dfLNA_musk, by = "OTU") %>% mutate(Lake = "Muskegon")
inland_df <- full_join(dfHNA_inland, dfLNA_inland, by = "OTU") %>% mutate(Lake = "Inland")
michigan_df <- full_join(dfHNA_michigan, dfLNA_michigan, by = "OTU") %>% mutate(Lake = "Michigan")
# Combine all of the three lakes together into one dataframe!
lake_dfscores <-
bind_rows(muskegon_df, inland_df, michigan_df) %>%
rename(HNA = RL.ranking.HNA, LNA = RL.ranking.LNA) %>%
gather(key = FCM_type, value = RL.ranking, HNA:LNA)
Load RL Score data
# Read in Data
dfHNA_musk_scores <- read.csv("Final/FS_Scores/Muskegon_fs_scores_HNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.09) %>%
mutate(Lake = "Muskegon")%>%
rename(OTU = X, RL.score.HNA = RL.score) %>%
dplyr::select(OTU, RL.score.HNA)
dfLNA_musk_scores <- read.csv("Final/FS_Scores/Muskegon_fs_scores_LNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.09) %>%
mutate(Lake = "Muskegon") %>%
rename(OTU = X, RL.score.LNA = RL.score) %>%
dplyr::select(OTU, RL.score.LNA)
dfHNA_inland_scores <- read.csv("Final/FS_Scores/Inland_fs_scores_HNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.13) %>%
mutate(Lake = "Inland")%>%
rename(OTU = X, RL.score.HNA = RL.score) %>%
dplyr::select(OTU, RL.score.HNA)
dfLNA_inland_scores <- read.csv("Final/FS_Scores/Inland_fs_scores_LNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.108) %>%
mutate(Lake = "Inland") %>%
rename(OTU = X, RL.score.LNA = RL.score) %>%
dplyr::select(OTU, RL.score.LNA)
dfHNA_michigan_scores <- read.csv("Final/FS_Scores/Michigan_fs_scores_HNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.248) %>%
mutate(Lake = "Michigan")%>%
rename(OTU = X, RL.score.HNA = RL.score) %>%
dplyr::select(OTU, RL.score.HNA)
dfLNA_michigan_scores <- read.csv("Final/FS_Scores/Michigan_fs_scores_LNA_5seq10.csv") %>%
dplyr::filter(RL.score > 0.286) %>%
mutate(Lake = "Michigan") %>%
rename(OTU = X, RL.score.LNA = RL.score) %>%
dplyr::select(OTU, RL.score.LNA)
# Combine HNA and LNA datasets together
muskegon_df_scores <- full_join(dfHNA_musk_scores, dfLNA_musk_scores, by = "OTU") %>% mutate(Lake = "Muskegon")
inland_df_scores <- full_join(dfHNA_inland_scores, dfLNA_inland_scores, by = "OTU") %>% mutate(Lake = "Inland")
michigan_df_scores <- full_join(dfHNA_michigan_scores, dfLNA_michigan_scores, by = "OTU") %>% mutate(Lake = "Michigan")
# Combine all of the three lakes together into one dataframe!
scores_RL_df <-
bind_rows(muskegon_df_scores, inland_df_scores, michigan_df_scores) %>%
rename(HNA = RL.score.HNA, LNA = RL.score.LNA) %>%
gather(key = FCM_type, value = RL.ranking, HNA:LNA)
Plot figure 4: RL Scores
# Reformat data to be matrix style for heatmapping
# HNA/LNA: Inland: 0.13/0.108, Michigan: 0.248/0.286, Muskegon: 0.09/0.09
musk_dat_scores <-
muskegon_df_scores %>%
dplyr::select(-Lake) %>%
dplyr::filter(RL.score.HNA > 0.09 | RL.score.LNA < 0.09) %>%
rename(Muskegon_HNA = RL.score.HNA,
Muskegon_LNA = RL.score.LNA)
inland_dat_scores <-
inland_df_scores %>%
dplyr::select(-Lake) %>%
dplyr::filter(RL.score.HNA > 0.13 | RL.score.LNA < 0.108) %>%
rename(Inland_HNA = RL.score.HNA,
Inland_LNA = RL.score.LNA)
michigan_dat_scores <-
michigan_df_scores %>%
dplyr::select(-Lake) %>%
dplyr::filter(RL.score.HNA > 0.248 | RL.score.LNA < 0.286) %>%
rename(Michigan_HNA = RL.score.HNA,
Michigan_LNA = RL.score.LNA)
matrix_RLscores <-
full_join(musk_dat_scores, inland_dat_scores, by = "OTU") %>%
full_join(michigan_dat_scores, by = "OTU") %>%
tibble::column_to_rownames(var = "OTU") %>%
as.matrix()
matrix_RLscores[is.na(matrix_RLscores)] <- 0
breakers <- seq(min(matrix_RLscores, na.rm = T), max(matrix_RLscores, na.rm = T), length.out = 21)
my_palette <- colorRampPalette(c("white","gray65", "red")) (n=20)
# Set the colors of the different columns going in
colz <- c("#FF933F", "#FF933F", "#EC4863", "#EC4863", "#5C2849", "#5C2849")
heatmap.2(matrix_RLscores,
distfun = function(x) dist(x, method = "euclidean"),
hclustfun = function(x) hclust(x,method = "complete"),
col = my_palette, breaks = breakers, trace = "none",
key = TRUE, symkey = FALSE, density.info = "none", srtCol = 30,
margins=c(5.5,7), cexRow = 1.25,cexCol = 1.25,
key.xlab = "RL Score",ColSideColors = colz,
lhei = c(1,9))
Top 10 from Fig 4 Scores
mich_top10 <- michigan_dat_scores %>%
top_n(., 10, Michigan_HNA)
inland_top10_HNA <- inland_dat_scores %>%
select(OTU, Inland_HNA) %>%
top_n(., 10, Inland_HNA)
inland_top10_LNA <- inland_dat_scores %>%
select(OTU, Inland_LNA) %>%
top_n(., 10, Inland_LNA)
musk_top10_HNA <- musk_dat_scores %>%
select(OTU, Muskegon_HNA) %>%
top_n(., 10, Muskegon_HNA)
musk_top10_LNA <- musk_dat_scores %>%
select(OTU, Muskegon_LNA) %>%
top_n(., 10, Muskegon_LNA)
df_top10 <- musk_top10_HNA %>%
full_join(., musk_top10_LNA, by = "OTU") %>%
full_join(., mich_top10, by = "OTU") %>%
full_join(., inland_top10_HNA, by = "OTU") %>%
full_join(., inland_top10_LNA, by = "OTU")
matrix_top10 <- df_top10 %>%
# Remove the zeros in the OTU names
mutate(OTU = str_replace(OTU, "00", "")) %>%
tibble::column_to_rownames(var = "OTU") %>%
as.matrix()
# The heatmap won't work with NAs
matrix_top10[is.na(matrix_top10)] <- 0
#png("Analysis_Figures/clustering_top10.png", units = "in", width = 6, height = 8, res = 300)
heatmap.2(matrix_top10,
distfun = function(x) dist(x, method = "euclidean"),
hclustfun = function(x) hclust(x,method = "complete"),
col = my_palette, breaks = breakers, trace = "none",
key = TRUE, symkey = FALSE, density.info = "none", srtCol = 30,
margins=c(5.5,7), cexRow = 1.25,cexCol = 1.25,
key.xlab = "RL Score",ColSideColors = colz,
lhei = c(1.5,9), key.title=NA)
#dev.off()
# WITHOUT the clustering!
heatmap.2(matrix_top10,
dendrogram = "none", Rowv = FALSE, Colv = FALSE,
col = my_palette, breaks = breakers, trace = "none",
key = TRUE, symkey = FALSE, density.info = "none", srtCol = 30,
margins=c(5.5,7), cexRow = 1.25,cexCol = 1.25,
key.xlab = "RL Score",ColSideColors = colz,
lhei = c(1,9), #lmat = rbind(1,2),
key.title=NA)
## Error in plot.new(): figure margins too large
ggplot heatmap
# In the order of OTUs without
df_top10 %>%
mutate(OTU = str_replace(OTU, "00", "")) %>%
gather(Lake_FCM, RL_score, Muskegon_HNA:Inland_LNA) %>%
ggplot(aes(x=Lake_FCM, y= fct_rev(OTU))) +
geom_tile(aes(fill=RL_score)) +
scale_fill_gradient2(na.value = "white", low = "grey", high = "red") +
theme(axis.title = element_blank(), axis.text.x = element_text(angle = 30, hjust = 1, vjust = 1))
Figure 5: Phylogenetic Tree
#### PHYLOGENETIC ANALYSIS
load("data/phyloseq.RData")
physeq.otu
## phyloseq-class experiment-level object
## otu_table() OTU Table: [ 13370 taxa and 859 samples ]
## tax_table() Taxonomy Table: [ 13370 taxa by 7 taxonomic ranks ]
otu_scores_df <- matrix_scores %>%
as.data.frame() %>%
tibble::rownames_to_column("OTU")
vector_of_otus <- as.vector(otu_scores_df$OTU)
physeq <- physeq.otu %>%
subset_taxa(., taxa_names(physeq.otu) %in% vector_of_otus)
# Which OTU is missing?
setdiff(sort(vector_of_otus), sort(taxa_names(physeq)))
## character(0)
setdiff(sort(taxa_names(physeq)), sort(vector_of_otus))
## character(0)
######################################### FASTTREE
fast_tree <- read.tree(file="data/fasttree/fasttree_newick_tree_HNALNA_rmN.tre")
fast_tree_tip_order <- data.frame(fast_tree$tip.label) %>%
rename(OTU = fast_tree.tip.label)
fasttree_physeq <- merge_phyloseq(physeq, phy_tree(fast_tree))
# Fix the taxonomy names
colnames(tax_table(fasttree_physeq)) <- c("Kingdom","Phylum","Class","Order","Family","Genus","Species")
###################################################################### ADD THE PROTEOBACTERIA TO THE PHYLA
phy <- data.frame(tax_table(fasttree_physeq))
Phylum <- as.character(phy$Phylum)
Class <- as.character(phy$Class)
for (i in 1:length(Phylum)){
if (Phylum[i] == "Proteobacteria"){
Phylum[i] <- Class[i] } }
phy$Phylum <- Phylum # Add the new phylum level data back to phy
t <- tax_table(as.matrix(phy))
tax_table(fasttree_physeq) <- t
fasttree_tax <- data.frame(tax_table(fasttree_physeq)) %>%
tibble::rownames_to_column(var = "OTU")
fasttree_df2 <- read.csv("data/fasttree/OTUnames_based_on_RLscores_MANUAL.csv") %>%
left_join(., fasttree_tax, by = "OTU") %>%
tibble::column_to_rownames("OTU") %>%
dplyr::select(Phylum)
## Let's root the tree
is.rooted(fast_tree)
## [1] FALSE
test_tree <- root(fast_tree, outgroup = "Otu001533", resolve.root = TRUE)
is.rooted(test_tree)
## [1] TRUE
phyfcm_fasttree_df3 <- read.csv("data/fasttree/OTUnames_based_on_RLscores_MANUAL.csv") %>%
left_join(., fasttree_tax, by = "OTU") %>%
tibble::column_to_rownames("OTU") %>%
dplyr::rename(FCM = fcm_type) %>%
dplyr::select(Phylum, FCM)
rooted_tree <-
ggplot(test_tree, aes(x, y)) + geom_tree() + theme_tree() +
geom_tiplab(size=3, align=TRUE, linesize=.5) #+
#geom_nodelab(vjust=-.5, size=3)
gheatmap(rooted_tree, phyfcm_fasttree_df3, offset = 0.1, width=0.3, font.size=3, colnames_angle=0, hjust=0.5) +
scale_fill_manual(values = phylum_colors,
breaks = c("Actinobacteria", "Alphaproteobacteria", "Bacteria_unclassified", "Bacteroidetes", "Betaproteobacteria",
"Cyanobacteria","Deltaproteobacteria", "Firmicutes", "Gammaproteobacteria","Omnitrophica", "Planctomycetes",
"Proteobacteria_unclassified", "unknown_unclassified", "Verrucomicrobia", "Both", "HNA", "LNA"))
